Qualcomm is teasing improvements to the Hexagon NPU in its upcoming Snapdragon 8 Elite Gen 6 Pro ahead of the Snapdragon Summit on September 22. The new NPU is built for agentic AI, featuring a transformer-focused Element Accelerator and a 50% larger shared memory that keeps frequently accessed model data close to the NPU. This design reduces memory bottlenecks, enabling agents to stay responsive while juggling longer context, more tools, and concurrent tasks.

The Element Accelerator is purpose-built for transformer workloads powering modern generative and agentic AI. Combined with scalar, vector, and matrix extensions, it accelerates operations critical for large models, helping agents respond faster and reason more efficiently without compromising mobile power efficiency. Qualcomm also claims up to 50% faster prefill for INT4 models.

The NPU is well suited to Mixture-of-Experts (MoE) models, which activate only a fraction of parameters per token. Overall, the new Hexagon NPU is designed for always-running AI, long-context reasoning, multimodal models, concurrent agents, and low-latency action loops.